82% of Shoppers Expect Brands to Make Good on Broken AI Promises

The new holiday report finds shoppers are open to AI-led purchases but expect brands to resolve problems and honour their promises.

Topics

  • More than six in ten shoppers (62%) are willing to let AI make purchases on their behalf, but they expect brands to take responsibility when the technology fails to deliver, according to Forethought AI Agents by Zendesk’s State of AI in CX 2026: Holiday Report.

    The research found that shoppers are increasingly open to using AI for tasks such as finding discounts, tracking prices and managing returns. However, their willingness to rely on the technology comes with clear expectations around accuracy, accountability, and problem resolution.

    Among the findings, 82% of shoppers expected brands to provide a refund or credit if their AI made a promise it failed to keep. The report also found that 41% had abandoned an online shopping cart because a chatbot could not answer a basic question about shipping or inventory.

    The findings point to a growing challenge for retailers: AI needs to do more than respond to customer queries. It must resolve issues, follow through on commitments and provide access to human support when necessary.

    The report comes as policymakers are also examining automated customer service. On October 20, New York City proposed rules that would require businesses to disclose when customers are interacting with automated systems and provide access to human representatives.

    “AI in customer service needs to do more than answer questions,” said Antoine Nasr, Head of AI at Forethought AI Agents by Zendesk. 

    “It needs to understand the customer’s issue and take action to resolve it. During the holidays, that could mean proactively addressing a delivery delay or processing a refund. That’s how AI earns shoppers’ trust.”

    Shoppers Want AI That Resolves Their Issues

    Forethought’s data shows that shoppers remain open to AI’s role in shopping, provided it delivers practical help when they need it.

    Difficulty reaching a human representative ranked as the leading customer service frustration in the company’s 2024 and 2025 surveys. This year’s findings show how unresolved interactions can affect purchasing decisions:

    • 41% had abandoned an online cart because a chatbot could not answer a basic question about shipping or inventory.
    • 72% had been frustrated by a bot that could not understand their issue.

    The findings suggest that ineffective automated support can directly affect sales, particularly during the holiday shopping season, when customers expect quick answers about stock availability, delivery, and orders.

    Shoppers Expect AI to Make Their Money and Time Go Further

    Shoppers also expect AI to deliver measurable value by reducing the time and money spent on holiday purchases.

    The report found that consumers were willing to use AI for specific shopping tasks, particularly those involving price comparisons and savings:

    • Shoppers expected AI to save them an average of $192 this season. More than half expected to save at least $100, while 12% expected savings of at least $500.
    • 57% trusted AI to track prices, while 42% were comfortable using it to find coupon codes.
    • 53% expected AI to save them at least an hour on holiday shopping, and 24% anticipated saving three hours or more.
    • Only 18% would allow AI to make purchases worth more than $500.

    The difference between these figures highlights the limits of consumer trust. While shoppers are comfortable delegating research and price tracking, they remain more cautious about giving AI control over higher-value transactions.

    For retailers, this means AI-powered shopping experiences need to demonstrate clear value while allowing customers to retain control over significant purchasing decisions.

    Brands Have to Stand Behind What Their AI Promises

    Shoppers are also placing responsibility for AI-generated commitments on the brands deploying the technology.

    The report found that 82% expected a refund or credit if a brand’s AI made a promise it failed to keep. Meanwhile, 15% said purchasing the wrong item based on a poor AI recommendation would be the biggest factor likely to ruin their holiday shopping experience.

    These expectations make accuracy and clear operational boundaries important considerations for businesses introducing AI into customer-facing services.

    “ A promise is a promise, whether it was made by a person or by AI,” said Erica Clayton, Head of Partnerships and CX at Forethought AI Agents by Zendesk. “The goal shouldn’t be for AI to never make mistakes. That’s not realistic. AI should only commit to what a business can actually deliver.”

    Clayton added that businesses should ground AI responses in approved policies, verify live order information before providing delivery dates and transfer complex issues to human representatives with the relevant customer context.

    Proactive AI Builds Loyalty

    Delivery delays are difficult to avoid during the holiday season, particularly when winter weather and increased carrier volumes affect fulfilment. However, the way brands communicate and respond to these disruptions can influence customer loyalty.

    Forethought’s research found that shoppers have limited patience when support is unavailable during peak periods:

    • 44% would wait less than 30 minutes for customer support before cancelling an order or shopping elsewhere.
    • 50% said they would be more loyal to a brand if its AI proactively flagged a delivery delay and offered a solution, such as free expedited shipping.

    The findings point to an opportunity for businesses to use AI beyond reactive customer service. By identifying potential problems and communicating solutions before customers need to ask, brands can make support more useful and reduce the likelihood of lost purchases.

    More Than Half of Shoppers Prefer AI-Guided Returns

    Returns are another area where AI could influence the customer experience, with shoppers showing a preference for convenient online processes.

    The report found that 55% preferred an instant AI-guided return process online rather than visiting a store or post office. It also found that 75% were at least somewhat likely to keep an item they were unhappy with if the return process was too complicated.

    The results suggest that a difficult returns experience can affect how customers evaluate a purchase and their willingness to shop with a brand again.

    “During peak season, when shoppers are buying more and have less time, a convenient return option can provide the confidence to complete a purchase—while a costly or complicated policy can send them elsewhere,” said Tim Fehr, Chief Growth Officer of Happy Returns.

    Key Findings From the State of AI in CX 2026 Report

    • Shoppers are open to AI-led purchases: 62% were willing to let AI make some purchases directly, although only 18% would trust it with transactions above $500.
    • Poor chatbot experiences affect sales: 41% had abandoned an online cart because a chatbot could not answer a basic question.
    • Consumers expect accountability: 82% expected brands to offer a refund or credit when AI failed to honour a promise.
    • Proactive support can build loyalty: 50% said they would be more loyal to brands whose AI flagged delivery delays and offered solutions.
    • AI-guided returns are gaining acceptance: 55% preferred an instant AI-guided online return process, while 75% were at least somewhat likely to keep an unwanted item if returning it was too complicated.

    The findings indicate that consumer acceptance of AI in customer experience will depend on whether businesses can translate automation into reliable service.

    As retailers introduce AI into shopping, support and returns, the expectation is shifting from simply providing automated interactions to taking responsibility for their outcomes. For brands, the ability to deliver on AI’s promises could become an important factor in retaining customer trust.

    ALSO READ: AI, Creativity, and the Risk of “Meh”

    Topics

    More Like This